3217统计中等essayshort
Why Optional Stopping Breaks a Fixed-Horizon p-Value
题目
An experiment is designed for a fixed horizon, but the desk checks the p-value every day and stops once it drops below 0.05. Why does that invalidate the nominal 5% Type I error guarantee? Why is the Bayesian posterior update answering a different question?
解题计时
0:00
提交作答时记录，用于后续平均用时统计。